My position on generative AI

What's your problem with AI?

To begin with, it never hurts to point out once again that “artificial intelligence” is above all a marketing term that in practice encompasses a variety of techniques ranging from simple algorithms (is a sorting program AI?) to LLM – _ChatGPT, Midjourney and so on: the latter are called “generative” AIs, because they don't involve processing data (as would an automatic image enhancement algorithm, or a text translator– just like the one I use to help me translate the blog into English), but generating text, images or whatever from a query (a prompt).

Generative AIs, by the very way they work, pose major problems:

• ecological problems, as they rely on disproportionate infrastructures (GPU farms, data centers) which cause an explosion in resource consumption (water¹, rare metals) and CO2 emissions², at a time when it would be urgent to reduce them. Of course, every digital application has an environmental impact, but that of generative AI is of a completely different order of magnitude;

• ethical problems, as they have been trained on the entirety of what the tech giants have been able to find on the Internet, which is akin to immense plundering in defiance of artists (see ³, ⁴ et ⁵). I'm all for free art, and have no great sympathy for copyright as it stands, but this headlong rush into generalized and unpunished plagiarism is diametrically opposed to my values, and quite frankly disgusting;

• democratic problems, because they industrialize large-scale standardization and Western (mainly American) cultural hegemony, with the inclusion of numerous biases (sexist, homophobic, racist, whatever you like), biases that exist in our society, but which here become automated, amplified and trivialized⁶.

Add to that the risks to our own brains: it has been shown that regularly using generative AI to delegate thinking and creative processes diminishes cognitive abilities and makes us less able to understand and comprehend the world around us⁷, not to mention cases of frankly frightening psychosis⁸...

For all these reasons (and others I'll detail below), I don't want to use generative AI.

Don't you think you're holding yourself back by not using AI?

Would generative AI make me more efficient in some areas? It might. But on what points? Writing ever more clichéd scenarios? Smooth out my style so that it looks like all the others?

And even then: I've already explained this in another article, but I don't necessarily want to be efficient. By that I mean that being efficient is not an end in itself. Even before the invention of generative AI, I could have been very efficient by keeping my old job as a well-paid engineer and outsourcing my designs to a draftsman or draftswoman in a poor country. But what's the point?

Obviously, I don't live in an ether disconnected from all material contingencies, and yes, I do have to earn a living from my art, but not at any price.

In fact, it may seem like a futile motivation, but the truth, the heart of the matter, is that... well, I love to write; I love to draw; I love to tell stories. In this, my vision is diametrically opposed to that of AI's cantors, starting with the CEO of Suno AI, a music generator, who declared:

It's not really fun to make music today. It takes a lot of time, a lot of practice, you have to get really good at an instrument or production software. I think most people don't appreciate the time they spend making music.⁹

A quote that will have given almost everyone who makes music a good laugh, I think. As a musician myself, I love every moment spent singing, playing the guitar or the piano... Yes, learning is long, sometimes difficult, but the pleasure you get from playing music, just like the pleasure you get from drawing things, telling stories... all that is irreplaceable, and certainly not by a stupid prompt.

Suno's CEO isn't an artist, he's a businessman, and as such, he's looking to optimize profit: yes, if I want to maximize my chances of making money from music, flooding Spotify with songs generated in 2 seconds by an AI might be a good option. Except that's not my goal.

My goal is to make art, and if possible to make a living from it. Besides, outside this blog, I play in small bands in my area, I don't earn a penny doing that, and I continue to do it for pure pleasure.

I'm not interested in making a living typing prompts, even though it would be so simple.

Don't you think you're just a drop in the bucket in a market that's about to become saturated with AI-generated art?

Oh yes, of course. But the cultural hegemony didn't wait for AI to flood the market. I'm going to quote myself (from this French radio column) again:

AI or not, mass culture is already mass-produced by algorithms: obviously, the Netflix scriptwriters who have a hyper-detailed manual with the right strings, the right recipes, the right plots to loop series after series, will we really see the difference if we replace them with AIs? AI does exactly what Netflix wants: to produce the statistically most expected result, the one that will satisfy the most people on average, on the chain, faster and faster.

In this respect, the advent of generative AI in art will threaten a whole host of professions, starting with those of the mass culture industry. But what about me? I wasn't part of it before. I'm both completely overwhelmed by this universe, and at the same time relatively indifferent and spared by it, because I don't play in the same league (for better or for worse).

AIs can already generate drawings of much higher quality than mine. In terms of scenarios/jokes, it's not quite there yet, but I imagine it's coming. Except that we're forgetting an essential point when we place ourselves outside the mass culture industry, in the indie world as we call it: the human relationship.

I regularly go to see concerts in my area, often local bands, not very well known, amateurs who do this between two days of salaried work. The music isn't the best in the world, the performers aren't always insanely talented, but that's not why I go: I go because I like seeing these people, because I like feeling what someone else is playing, because I like the relationship we have when these people are playing on stage and I'm in the audience (and vice versa, since as I said above, I sometimes play too).

Yes, AIs already make better music, better crafted, better written, better “recorded”. But really, I don't care: I don't listen to musicians with the aim of finding the best music in the world, as I would optimize a route with Waze. I listen to them because I want to hear what people have to say, to tell me, to see what their human experiences have inspired in terms of lyrics, melodies...

By the very essence of the process, an AI won't be able to replace that. It can imitate a whole lot of stuff, be more efficient and optimized, but at the end of the day, a human being who appreciates the artistic work of another because it's a relationship that has a meaning that goes beyond the work itself, that can't be generated.

Maybe I'm being naive, but I can already see how much AI-based art is hated, and not just by artists who are legitimately afraid for their future.

So yes, a whole lot of people will consume AI art the way we consume junk food: maybe I will too one day, I don't claim to be above the fray. After all, I sometimes go to fast-food restaurants. But I dare to believe that the people who follow me now, and who will follow me one day, don't just follow me for my pretty drawings, but also because these drawings come from me, from a specific person, who has a story as singular as theirs, someone of flesh and blood with whom you can talk, whom you can even meet... and that all this means something; that it's not soluble in the giant statistics of the standardizing machines that are generative AIs; that it's a human relationship that goes beyond drawn scribbles or written jokes.

In short: I know that the junk food of AI art will soon be everywhere. But I welcome you to my little neighborhood restaurant with old-fashioned, human art.

Wishing you bon appétit!